An Analysis of the Calque Phenomena Based on Comparable Corpora

In this short paper we show how Comparable corpora can be constructed in order to analyze the notion of ’calque’. We then investigate the way comparable corpora contribute to a better linguistic analysis of the calque effect and how it can help improve error correction for non-native language productions. 1 Aims and Situation Non-native speakers of a language (called the target language) producing documents in that language (e.g. French authors like us writing in English) often encounter lexical, grammatical and stylistic difficulties that make their texts difficult to understand. As a result, the professionalism and the credibility of these texts is often affected. Our main aim is to develop procedures for the correction of those errors which cannot (and will not in the near future) be treated by the most advanced text processing systems such as those proposed in the Office Suite, OpenOffice and the like, or advanced writing assistance tools like Antidote. In contrast with tutoring systems, we want to leave decisions as to the proper corrections up to the writer, providing him/her with arguments for and against a given correction in case several corrections are possible. To achieve these aims we need to produce a model of the cognitive strategies deployed by human experts (e.g. translators correcting texts, teachers) when they detect and correct errors. Our observations show that it is not a simple and straightforward strategy, but that error diagnosis and corrections are often based on a complex analytical and decisional process. Most errors result from a lack of knowledge of the target language. A very frequent strategy for authors is to imitate the constructions of their native language so that the production resembles standard terms and constructions of the target language. This approach based on analogy is called a calque when surface forms are taken into consideration (Hammadou, 2000), (Vinay et al. 1963). The errors produced in this context may be quite complex to characterize, and they are often difficult to understand. When attempting to correct these errors, we find it interesting to have access to some of the characteristics of the native language of the author so that a kind of ’retro-analysis’ of the error can be carried out. This would allow a much better rate of successful corrections, even on apparently complex errors involving long segments of words in a sentence. Works on the correction of grammatical errors made by human authors (e.g. Writer’s v. 8.2) have recently started to appear. These systems do not propose any explicit analysis of the errors nor do they help the user to understand them. The approach presented here, which is still preliminary, is an attempt to include some didactic aspects into the correction by explaining to the user the nature of her/his errors, whether grammatical or stylistic, while weighing the pros and cons of a correction, via argumentation and decision theories (Boutiler et ali. 1999), (Amgoud et ali. 2008). Persuasion aspects are also important within the didactical perspective (e.g. Persuation Technology symposiums), (Prakken 2006). Finally, the calque (direct copy) effect has been studied in the didactics of language learning, but has never received much attention in the framework of error correction, where a precise analysis of its facets needs to be conducted. In this short document we present the premises of an approach to correcting complex grammatical and lexical errors based on an analysis of the calque effect. Calque effects cannot easily be reduced to the violation of a few grammar rules of the target language: they need an analysis of their